The distinction among electromagnetic radiation source models based on directivity with support vector machines

Directivity, as a characteristic parameter of electromagnetic radiation source, could be used to classify different radiation sources, and the ESD events could be considered as typical sources. The parameter could be measured by electric field intensity radiating in all directions in the space. In this paper, we would build three basic antenna models which are all working at 3 GHz and set cube receiving arrays along the main lobe of their radiation patterns to receive the data of far field electric intensity in groups. Then the SVM method is adopted to analyze training data set, build and test the classification model. The classification result would be compared with that of BP Neural Network method. Owing to the powerful nonlinear simulation ability, the SVM method gets higher classification accuracy in noise environment. At last, the classification model is comprehensively evaluated in three aspects, which are capability of noise immunity, f1 measure and the normalization method.